Just a brain exercise:
AI learns from meatbags.
We're told that us healthy people are in the minority, and mentally ill people, I mean LGBT are the majority.

So if AI always gets so racist, sexist, transphobe, antisemite, or whatever else, who do you think the majority REALLY is?
Is it because AI is systematically racist bigot?
Or is it that LGBT isn't really as big as we're all told it to be after all?